About two weeks ago, I mentioned that one could use web usage statistics to monitor the penetration of Windows 10. As Microsoft itself mentioned, Windows 10 had a very successful launch and huge numbers of installations during the first days after launch.
Here I would like to show how web usage of Windows 10 looks after 20 days and how it compares to previous Windows launches.
In the following, I will use data from StatCounter and focus primarily on US data. I do not consider the way StatCounter aggregates global data to be meaningful so I will only look at data per country (I don’t agree with NetMarketShare’s weighting method either).
The following the US StatCounter data for only PCs (excluding tablets, smartphones and consoles). The data goes back to 2008 so we can monitor the launch of Windows 7, as well as Windows 8 and 8.1. Windows 10 does not yet have its own line, but from the CSV data that you can also download, we know that the final uptick in the “other” category at the right end of the graph is almost entirely due to Windows 10, which garnered 5.72% web usage share on August 2015 (to date).
We can clearly observe the following;
- Windows 7, 8 and 8.1 experienced stable increases in usage share up until the point where a new version of Windows was released. Although there is a slight flattening of the slope, this is to be expected as the number of computers with older versions decrease.
- The only event that significantly slows the increase of usage of a certain Windows version, is the release of a newer version. Growth of Windows 7 web usage flattened only when Windows 8 was released, and likewise, Windows 8 growth slowed only after Windows 8.1 was released.
- Importantly, there is little evidence that there is an initial burst of upgrades followed by a severe slowing down.
- However, the rate of increase in web usage differs significantly between versions. Windows 7 had much better uptake compared to Windows 8. Windows 8.1 improved on Windows 8, but was still not as rapid as Windows 7. This suggests that uptake is positively related to how well the OS is received, and not so much to the sales of PCs in general.
- Windows 7 managed to penetrate 50% of PC web usage in August 2012, 3 years after being launched in October 22, 2009. Web usage share in the first month after launch rose to 6.61% maximum.
- Windows 10 web usage increase in the first month following launch is phenomenally rapid. It has risen to a maximum of 7.97% in the first 20 days since launch.
There are some questions that arise from this data;
- How could Windows 7 achieve 50% penetration in 3 years when the refresh cycle is ~5 years?: It is widely known that most Windows OS upgrades come from customers purchasing new PCs. Most Windows customers do not go out and buy new versions of the OS and install it themselves. The way to understand this is that inside the 1.5 billion PC installed base, there are some PCs that are used frequently and some that are only seldom used. The seldom used PCs do not get upgraded or refreshed but they also do not show up in web usage data. Hence if we look only at the frequently used PCs which do show up in web usage data, the refresh cycle will probably be closer to 3 years than 5 years.
- Can we estimate the size of pent-up demand?: Windows 7 managed to reach 50% penetration in 3 years. On the other hand, even after 3 years since Windows 8, the combined web usage share of Windows 8 and 8.1 is only 20%. This suggests that there might be 30% points of pent-up demand.
- Can Windows 10 maintain its current rapid adoption rate?: This is the big question that needs to be addressed. At this point everything is highly speculative, but I think the signs are positive. I do not think it is unreasonable to expect 40% web usage within a year (Windows 7 reached almost 25%).
A rapid upgrade towards Windows 10 is a pre-requisite for Microsoft’s strategy of moving towards a universal platform for multiple devices and getting developers on-board. Web usages statistics suggest that this is what we are seeing right now.
Naofumi Kagami 